A fast new method for segment identification in water distribution systems

This paper presents a fast new method for the identification of the segments formed after the installation and closure of isolation valves in water distribution systems. The reduced computing times of the method make it suitable for the analysis of real networks, which always comprise a large number of nodes and pipes. Unlike other methods based on the use of topological matrices, in which the structure of the original network is modified by inserting fictitious pipes in order to represent the presence of the valves, the method presented in this paper uses the original topological structure of the network and two auxiliary matrices to contain valve-related information. This makes it possible to limit significantly the size of topological matrix used for segment identification and thus to reduce computational times. The method proposed can be used in the case of an existing network (where isolation valves are already present) in order to identify segments which are formed after these valves have been closed; it can also be applied to the case of a network which has to be designed or restored, where valve optimal positions have to be set. In this second case the methodology is coupled with a multi-objective genetic algorithm in which the objective functions to optimize are the valve number or cost (to minimize) and the system reliability in terms of maximum (or weighted average) unsupplied water demand after the isolation of segments (to minimize). The numerical applications of the method concern a complex real water distribution network. In particular, firstly the segments formed after the closure of isolation valves really present in the network are identified and analysed. Secondly, the methodology is used in an optimization context, providing solutions that are more cost effective than the present valve configuration for fixed system reliability.